High-resolution direction of arrival estimation method based on optimal subspace fitting under non-Euclidean metrics

To address the issue of insufficient spatial resolution in direction of arrival (DOA) estimation under limited physical array aperture conditions, this paper proposes a high-resolution DOA estimation method based on optimal subspace fitting with non-Euclidean metrics. By establishing the mapping relationship between the geometric distance of matrices and signal direction, a DOA estimation framework is constructed based on covariance matrix fitting. Introducing matrix information geometry theory, the method replaces traditional Euclidean distance metrics with non-Euclidean distance metrics, and proposes to utilize an improved symmetric Kullback-Leibler divergence to accurately measure the similarity between matrices. The optimal subspace estimation algorithm is employed to enhance the accuracy of signal subspace estimation, and high-resolution DOA estimation is achieved through the fitting between the theoretical signal subspace and the estimated optimal signal subspace. Simulation and experimental results demonstrate that the proposed method outperforms traditional high-resolution DOA estimation methods in both resolution capability and estimation accuracy, while maintaining stable performance under limited array aperture and number of snapshots, and exhibits robustness in complex underwater environments with signal power differences.